Measuring and improving the innovative potential of research and translation through temporal knowledge graph embedding

University Of Michigan At Ann Arbor, Ann Arbor MI

Investigators

Abstract

Project Summary Biomedical research relies on complex, evolving, multidisciplinary collaborations driven by researchers who spend entire careers working at the moving frontiers of knowledge. The resulting innovations contribute dramatically to human health, wealth and well-being. But increasing evidence suggests that marquee successes are becoming more difficult and costly to achieve, a phenomenon commonly attributed to substantial increases in the scale, scope, velocity and competitiveness of science over several decades. This project (1) develops flexible computational, data and network science tools (2) integrated by a social scientific framework to (3) analyze discovery and innovation dynamics and (4) improve individual and collective capacities to identify and realize solutions for even wildly challenging problems. This use-inspired approach is applicable across science, technology, engineering and mathematics (STEM) fields. Findings will thus contribute to knowledge in network science, science of science, sociology and economics of science and innovation and related fields while simultaneously offering new possibilities for application to the long, complex, uncertain processes that characterize biomedical research and development (R&D). By employing advanced techniques from network science, information theory, and machine learning, we aim to identify and predict emerging opportunities for groundbreaking research—termed 'adjacent possibles'—and accelerate transformative breakthroughs in biomedical fields and in science policies designed to improve human health through research, innovation, and translation. We examine the evolving social and conceptual landscape of research and its implications for innovative possibility, the knowledge and collaborative dynamics of teams, their capacity to achieve novel research aims and the sources and trajectories that yield long-term high impact disruptive findings. Results will increase understanding of how biomedical R&D can more effectively harness diverse knowledge to achieve disruptive scientific outcomes in critical areas such as mental health, addiction, and aging-related diseases. Project outcomes will inform strategies to foster transformative biomedical research, identify and enhance trajectories for high-impact translation, and thus contribute to human health. The resulting models and tools will increase capabilities for identifying emerging research areas and assessing risk in research to help inform policy, funder and investigator decision-making. This project directly addresses the National Library of Medicine's call for innovative approaches to advance biomedical informatics and data science, potentially revolutionizing our understanding of scientific innovation dynamics and informing evidence-based science policy decisions.